Separate-and-Detect — Audio Demo Page

Live page: https://ddman1101.github.io/dssdtm.github.io/

This demo page showcases our separation-then-transcription pipeline for automatic drum transcription (ADT). It contains:

• Abstract: a short overview of the task and method.
• Pianoroll + Six‑track Audios (Two examples):
  • For each example: one pianoroll image.
  • Four 6‑audio sets per example: Target, Prediction—ADTOF, Prediction—LarsNet, Prediction—MSG‑LD (+OB & TB).
  • Each set includes Mixture + the five stems above.

Data note

The drum sounds in our examples are rendered from one-shot samples extracted from the StemGMD dataset.